Lamb Shift in Atomic Hydrogen - University of...

26
Lamb Shift in Atomic Hydrogen 1 Background The spectrum of the hydrogen atom has proved to be the Rosetta stone of modern physics: once this pattern of lines had been deciphered much else could be understood. —Arthur Schawlow The goal in this experiment is to measure the Lamb shift in the Balmer α transition of atomic hydrogen. The Lamb shift cannot be explained by the Schr¨ odinger or Dirac formulations of quantum mechanics. It can, however, be explained by a theory known as quantum electrodynamics—a theory whose development was intimately linked to experimental observation of the Lamb shift. Note: Reading the article from Scientific American, “The spectrum of atomic hydrogen,” by ansch, Schawlow and Series [1] (through p. 107) is strongly recommended. A copy is available for students to download from the class website. Although we do not use a dye laser, that section is still relevant reading as our tunable diode laser serves the same function as the dye laser described in the article. Also, this write-up presents a number of exercises to the reader. As part of this experiment you are required to complete all exercises. Most can be done before doing the lab. 1.1 Classical description of the hydrogen atom The spectrum of the hydrogen atom was the first to be described quantitatively and modeled from first principles. In 1885 Balmer discovered that the wavelengths of the then known lines in the hydrogen spectrum were described by the formula λ = 3646 n 2 n 2 - 4 ! ˚ A (1 ˚ A = 10 -10 m) (1) where n is an integer and takes on the values 3, 4, 5,... This series of lines is now known as the Balmer series, and it is the line corresponding to n = 3, known as the Balmer α line, that we will investigate in this experiment. As an aside, atomic transitions were first studied in the optical (humanly visible) portion of the electromagnetic spectrum. Transitions were observed using an instrument call a spectrograph which separates the incoming light by wavelength. Transitions at different wavelengths appear as distinct vertical lines in the spectrograph view field, hence the term “line” to denote a transition at a particular wavelength. Exercise 1 Use Balmer’s formula to calculate the wavelength for the Balmer α line. In 1890 Rydberg discovered a more general form of Balmer’s formula which, when applied to the hydrogen spectrum has the form 1 λ = R H 1 m 2 - 1 n 2 (2) where R H is a constant, known as the Rydberg constant (R takes on slightly different values depending on the mass of the atom under investigation), and m and n are integers with m<n. In 1

Transcript of Lamb Shift in Atomic Hydrogen - University of...

Lamb Shift in Atomic Hydrogen

1 Background

The spectrum of the hydrogen atom has proved to be the Rosetta stone of modern physics:once this pattern of lines had been deciphered much else could be understood.

—Arthur Schawlow

The goal in this experiment is to measure the Lamb shift in the Balmer α transition of atomichydrogen. The Lamb shift cannot be explained by the Schrodinger or Dirac formulations of quantummechanics. It can, however, be explained by a theory known as quantum electrodynamics—a theorywhose development was intimately linked to experimental observation of the Lamb shift.

Note: Reading the article from Scientific American, “The spectrum of atomic hydrogen,” byHansch, Schawlow and Series [1] (through p. 107) is strongly recommended. A copy is available forstudents to download from the class website. Although we do not use a dye laser, that section isstill relevant reading as our tunable diode laser serves the same function as the dye laser describedin the article.

Also, this write-up presents a number of exercises to the reader. As part of this experiment youare required to complete all exercises. Most can be done before doing the lab.

1.1 Classical description of the hydrogen atom

The spectrum of the hydrogen atom was the first to be described quantitatively and modeled fromfirst principles. In 1885 Balmer discovered that the wavelengths of the then known lines in thehydrogen spectrum were described by the formula

λ = 3646

(

n2

n2 − 4

)

A (1A = 10−10m) (1)

where n is an integer and takes on the values 3, 4, 5, . . . This series of lines is now known as theBalmer series, and it is the line corresponding to n = 3, known as the Balmer α line, that we willinvestigate in this experiment. As an aside, atomic transitions were first studied in the optical(humanly visible) portion of the electromagnetic spectrum. Transitions were observed using aninstrument call a spectrograph which separates the incoming light by wavelength. Transitions atdifferent wavelengths appear as distinct vertical lines in the spectrograph view field, hence the term“line” to denote a transition at a particular wavelength.

Exercise 1 Use Balmer’s formula to calculate the wavelength for the Balmer α line.

In 1890 Rydberg discovered a more general form of Balmer’s formula which, when applied to thehydrogen spectrum has the form

1

λ= RH

(

1

m2−

1

n2

)

(2)

where RH is a constant, known as the Rydberg constant (R takes on slightly different valuesdepending on the mass of the atom under investigation), and m and n are integers with m < n. In

1

1913 Bohr proposed a model of the atom in which the (orbital) angular momentum of the electronis quantized in units of h. Using this model he derived an expression for the energy levels En ofthe hydrogen atom

En = −

(

mZ2e4

(4πε0)22h2

)

1

n2(3)

where m is the mass of electron, Z is the number of protons in the nucleus (Z = 1 for hydrogen),e is the charge of the electron and nh is the orbital angular momentum, with n = 1, 2, 3, . . .A transition from a state of higher energy Eni

to one of lower energy Enfshould result in the

emission of radiation with energy

E = Eni− Enf

=

(

mZ2e4

(4πε0)22h2

)[

1

n2f

−1

n2i

]

(4)

where nih and nf h are the electron orbital angular momenta of the initial and final states, re-spectively. Comparing this formula with the Rydberg formula and using the relation for radiatedenergy, E = hν = hc/λ, we find that RH should be

RH =me4

8ε20h

3c. (5)

Refined spectroscopic measurements indicated that this expression was not quite correct—a situa-tion that Bohr remedied by replacing the electron mass with the reduced mass µ of the atom,

µ =mM

M + m(6)

where M is the mass of the nucleus.

But other problems remained. In 1887 Michelson, using his interferometer to investigate the shapeof spectral lines, discovered that the Balmer α line consisted of not a single line but of two linesseparated by a fraction of an angstrom. Bohr’s simple model of the atom had no means of accountingfor this feature.

In 1916 Arnold Sommerfeld presented a model of the hydrogen atom allowing for the possibility ofelliptical in addition to strictly circular electron orbits. Applying relativistic corrections to theseelliptical orbits, Sommerfeld’s model predicted the Balmer α line to consist of more than one com-ponent, just as Michelson had observed. But, being an ad hoc combination of classical and quantumphysics, the Sommerfeld model was restricted in scope and left much to be desired. Quantitativelyit was unable to account for features such as magnetic effects or the different intensities of thecomponents of the Balmer α line.

1.2 Quantum mechanical description of the hydrogen atom

A more fundamental treatment of the hydrogen spectrum, as well as the spectra of other atoms,awaited the development of quantum mechanics in the 1920’s. In 1928 Dirac presented a relativisticquantum theory with predicted energy levels for the hydrogen atom [2, pp. 284–6]:

Enj = En

[

1 +

(

α

n

)2 ( n

j + 1/2−

3

4

)

]

(7)

2

where

En = −

(

µe4

(4πε0)22h2

)

1

n2. (8)

The integer n is known as the principal quantum number, and as before can take on the valuesn = 1, 2, 3, . . . The α in Eq. (7) is a dimensionless constant known as the “fine-structure constant”;it represents the scale by which the energy corrections are reduced relative to the main energy termEn that is due to the Coulomb (electrostatic) interaction between the proton and electron. Thefine structure constant is a combination of various fundamental constants:

α =e2

4πε0hc≈

1

137. (9)

The number j in Eq. (7) represents the total angular momentum of the electron; j is a combinationof the orbital angular momentum and the spin angular momentum, and can take on the valuesj = 1/2, 3/2, . . . , n − 1/2. Orbital angular momentum is specified by the integer l, which can takeon the values 0, 1, . . . , n − 1. Spin angular momentum is specified by the number s; for a singleelectron, s = 1/2. The total angular momentum number j can take on the values l ± s (l ± 1/2 fora single electron) depending on whether the angular momenta are parallel or antiparallel. Stateswith the same n but different values of j are called “sublevels.”

An aside about notation of atomic states: states are specified using a long established notation,n2s+1Xj , where n is the principle quantum number, s is the spin quantum number, X denotes theorbital angular momentum number (S for l = 0, P for l = 1, D for l = 2, F for l = 3), and jdenotes the total angular momentum quantum number. A state with n = 3, s = 1/2, l = 2 andj = 3/2 is denoted as 32D3/2. For atomic hydrogen s is always 1/2 and so the 2s + 1 term is oftendropped from the specification, 32D3/2 becoming 3D3/2, for example.

The Dirac formulation accounts for energy shifts due to relativistic corrections and magnetic in-teractions between the orbital and spin magnetic moments. This equation tells us that the energylevels depend only on the principal quantum number n and the total angular momentum j, i.e.,states (for a given n) with the same value of j will have the same energy. For example, a statewith l = 1 and j = 1/2 (orbital and spin angular momenta antiparallel), designated P1/2, shouldhave the same energy as a state with l = 0 (no orbital angular momentum), designated S1/2, andj = s = 1/2.

When applied to the Balmer α transition (the transition under investigation in this experiment)where ni = 3 and nf = 2, the Dirac equation predicts seven allowed transitions between thevarious sublevels of the n = 3 and n = 2 levels. Selection rules, which govern how much theangular momentum can change in a transition, prohibit transitions between certain sublevels. Theselection rules applying to the type of transitions (electric dipole) under consideration here are:∆l = ±1, ∆j = 0,±1.

Exercise 2 Use the selection rules and the notation described above to list all seven possible tran-sitions between the n = 2 and n = 3 states in atomic hydrogen. Which transitions does the Diractheory predict to have the same energy?

Figure 1 shows the seven transitions, with the height of the solid lines in the lower part of thefigure representing the strength of the transition. The curve above the seven lines represents theDoppler broadened profile that is observed when looking at the Balmer α radiation coming from a

3

Figure 1: Doppler broadened profile of Balmer α lines.

hydrogen discharge lamp. The Doppler principle states that light emitted by an atom moving withvelocity v relative to the observer, will, according to the observer, have frequency

ν = ν0 (1 ± v/c) (10)

where ν0 is the frequency of the light emitted by an atom at rest relative to the observer and c isthe speed of light. The + sign is associated with the atom moving toward the observed, and the −

sign with motion away from the observer.

Doppler broadening is due to the thermal motion of the atoms. The full width half maximum(FWHM) due to Doppler broadening is given by the formula

∆ν =2ν0

c

2kBT ln 2/M (11)

where kB is Boltzmann’s constant, T is the absolute temperature, M is the mass of the atom, andν0 is the center frequency of the transition [3, p. 189]. In the absence of Doppler broadening, thetransitions of interest in this experiment have linewidths ∼ 100 MHz, considerably smaller thanthe Doppler broadened linewidth. (Linewidths observed in the absence of Doppler broadening andother broadening mechanisms, e.g. pressure broadening, are called natural linewidths, and are dueto the finite lifetime of the excited state. The natural linewidth is determined by the uncertaintyprinciple, ∆E∆t ≥ h, where ∆t is the excited state lifetime).

Note: Throughout this write-up energy is specified in terms of frequency, which is obtained bydividing the actual energy by Planck’s constant, i.e., ν = E/h.

Exercise 3 Calculate the FWHM due to Doppler broadening for the hydrogen Balmer α transitionat 300 K, and compare it to the energy difference between the 3D3/2 and 3P1/2 states.

4

1.3 Experimental results that pointed the way

At the time Dirac presented his theory, experimenters had observed only two distinct Balmer αlines. Doppler broadening was clearly a limiting factor in attempting to resolve additional lines.What are the options for addressing this problem? The FWHM formula tells us that decreasingthe temperature or increasing the mass will reduce the FWHM. In 1938 R. Williams exploited bothof these factors by measuring the Balmer α spectrum from a deuterium discharge lamp cooled to100 K [4]. He succeeded in resolving for the first time a third Balmer α component, the componentcorresponding to the 3P1/2 → 2S1/2 transition. Figure 2 shows a plot of Williams’ results for the

Figure 2: Intensity curves for Dα and Hα lines from the experiment of Williams (1938).

Balmer α lines of both deuterium (Dα) and hydrogen (Hα). (In this case, energy units are in cm−1,or wavelengths per cm. The value in cm−1 is often called the wavenumber. The conversion betweenenergy stated in wavenumber and energy stated in frequency is 1 cm−1 = 29.979 GHz. To find theactual energy, multiply the energy stated in cm−1 by hc.)

Exercise 4 Calculate the FWHM due to Doppler broadening for the Balmer α transition in deu-terium at 100 K, and compare it to the FWHM you calculated above.

The Dirac theory predicted the separation between the 3D3/2 → 2P1/2 and 3P1/2 → 2S1/2 tran-sitions to be just the energy difference between the 3D3/2 and 3P1/2 states (remember, the 2S1/2

and 2P1/2 states are predicted to have the same energy since they have the same value of j). Thepredicted separation is 3.24 GHz (the 3D3/2 state lies above 3P1/2 state).

But Williams’ measurements showed this separation to be considerably larger. In Williams’ words:“The most striking aspect of the data . . . is the large percent deviation from theory in the intervalfrom component 2 to component 3.” [4] (Component 2 is the 3D3/2 → 2P1/2 transition andcomponent 3 is the 3P1/2 → 2S1/2 transition.) How could this discrepancy be explained? Was itpossible that the 2S1/2 and 2P1/2 states were not degenerate? (Degeneracy means having the sameenergy.)

Exercise 5 From the graph of Williams’ deuterium data in Fig. 2, estimate the energy differencebetween the 3D3/2 → 2P1/2 and 3P1/2 → 2S1/2 transitions.

In 1947 W. E. Lamb and R. C. Retherford performed a brilliant experiment in which they directlymeasured the energy difference between the 2S1/2 and 2P1/2 states [5]. In their experiment (see

5

Figure 3: Configuration of the Lamb-Retherford experiment (1947).

Fig. 3), they prepared a beam of hydrogen atoms in the 1S1/2 ground state. The beam is bombardedwith electrons with the result that some of the atoms are excited to the 2S1/2 state. Opticaltransitions from this state that go directly to the 1S1/2 ground state are prohibited by selectionrules (which one?) resulting in a very long (∼ 0.1 second) lifetime for the 2S1/2 state [3, p. 230].The atoms then pass through a region where they are exposed to electromagnetic radiation withenergy equal to the energy difference between the 2S1/2 and 2P1/2 states. The radiation has theeffect of inducing transitions between these states, and once an atom is in the 2P1/2 state it decaysto the ground state with a lifetime of 10−9 seconds.

After passing through the region with electromagnetic radiation, the atoms strike a tungsten foil.Upon striking the foil, atoms still in the 2S1/2 state decay to the ground state and in so doing liberateelectrons from the foil in a process called Auger emission. By measuring the emission current fromthe foil with the electromagnetic radiation turned on (fewer electrons liberated) and with theradiation turned off (more electrons liberated), Lamb and Retherford were able to determine theenergy difference between the 2S1/2 and 2P1/2 states. Their initial measurements indicated thisdifference to be about 1000 MHz.

Exercise 6 Why was Doppler broadening not a problem in the measurements of Lamb and Rether-ford?

1.4 QED picture of the hydrogen atom

The importance of the Lamb-Retherford measurement cannot be overstated—it was a milestonein experimental physics which spurred major developments in theoretical physics. In Dirac’s ownwords: “No progress was made for 20 years. Then a development came, initiated by Lamb’s discov-ery and explanation of the Lamb shift, which fundamentally changed the character of theoreticalphysics.”

The Lamb-Retherford measurement led to the development of a theory known as quantum elec-trodynamics (QED). This theory proposes that the electromagnetic field consists of a collection ofquantized harmonic oscillators, with the presence of a photon of energy hν corresponding to theoscillator with frequency ν being in the first excited state. (A harmonic oscillator is one with aparabolic potential energy well.) A fundamental property of the quantized harmonic oscillator isthat the energy in the ground state is not zero, but hν/2 [2, pp. 222–3]. This energy is called thezero-point energy. One way of thinking about the zero-point energy is that it is a requirement ofthe uncertainty principle: a particle in a potential well with zero kinetic energy would have both a

6

well defined position (bottom of the well) and a well defined momentum (p = 0). This is forbiddenby the uncertainty principle which requires that ∆x∆p ≥ h.

Even in the absence of any external radiation, non-zero electromagnetic fields are present due to thefields associated with zero-point energy. These fields are known as “vacuum fluctuations,” so namedbecause they exist even in the absence, or “vacuum” of observable photons. The electromagneticfields associated with vacuum fluctuations act on the electron to smear out its position relative tothe nucleus. (The proton with its greater mass is less affected by vacuum fluctuations.) Because ofthis smearing, the electron charge is distributed over a larger volume of space, and less of its chargeexperiences the strongest Coulomb potential closest to the nucleus. The result is a reduction in thebinding energy of the electron to the nucleus. Electrons in S states (l = 0) have a greater probabilityof being close to the nucleus than those in P or D states and so experience a greater reduction inbinding energy than is the case for P or D states. The major effect of the vacuum fluctuations is toshift S states upward in energy relative to P states. QED calculations show that the contributionto the energy difference between the 2S1/2 and 2P1/2 states due to vacuum fluctuations is +1086MHz. Calculations not based on formal QED theory, but still in fairly good quantitative agreementwith the Lamb-Retherford measurement, were made shortly after experimental results were firstannounced. The calculation by Welton is straightforward and the interested reader can pursue thisreference to get a feel for a simple theoretical treatment of the Lamb shift [6].

Exercise 7 Explain how the upward shift in S state energy relative to a P state energy accountsfor the energy difference observed by Williams (between the 3D3/2 → 2P1/2 and 3P1/2 → 2S1/2

transitions) being larger than predicted by the Dirac theory.

A second QED phenomenon contributing to the energy shift between S states and P states isknown as vacuum polarization. Virtual pairs of electrons and positrons are constantly created andannihilated in the Coulomb field of the proton. Being of opposite charge, virtual electrons areattracted to the proton, while positrons are repelled. The result is that the charge of the protonis partially screened by the virtual electrons. Electrons (ones bound to protons) penetrating thecloud of virtual particles will experience a stronger attraction to the nucleus than predicted by theCoulomb (1/r) potential, and hence will be more tightly bound to the nucleus. Electrons in Sstates (l = 0) have a greater probability of being close to the nucleus than those in P or D statesand so experience a greater increase in binding energy than is the case for P or D states. QEDcalculations show that the contribution to the energy difference between the 2S1/2 and 2P1/2 statesdue to vacuum polarization is −27 MHz. Note that this energy shift is in opposite direction fromthat due to vacuum fluctuations. Other smaller effects such as reduced mass and finite size of theproton also contribute to the Lamb shift. The sum of these smaller contributions is approximately−1 MHz. Recent calculations for all effects contributing to the Lamb shift result in a value

1057.833 (4) MHz [7].

The best experimental value for the 2S1/2 − 2P1/2 energy difference is

1057.845 (3) MHz [7].

That these values are so close is an outstanding triumph of both experimental and theoreticalphysics.

7

1.5 Some final details

The astute observer will have noticed that the energy level diagram at the top of Fig. 1 alreadyshows the splitting of the 2S1/2 and 2P1/2 states due to the Lamb shift. Indeed, it is this splittingthat results in 7 distinct lines in the Balmer α transition instead of the 5 predicted by the Diracequation. But what about the other splittings shown in the diagram—the splitting between the3S1/2 and 3P1/2 states and that between the 3P3/2 and 3D3/2 states? What is the origin of theseenergy differences? It turns out these splittings are also manifestations of the Lamb shift. Thesame mechanisms, primarily vacuum fluctuations and vacuum polarization, that account for thesplitting of the 2S1/2 and 2P1/2 states result in the splittings of the n = 3 states having the samevalue of total angular momentum j.

Exercise 8 The magnitude of the Lamb shift goes as n−3. Given a Lamb shift of 1058 MHz betweenthe 2S1/2 and 2P1/2 states, predict the spitting between the 3S1/2 and 3P1/2 states. Compare yourprediction to the measured value of 315 MHz.

Figure 4 is a repeat of Fig. 1, but in addition shows an example of data similar to that you willacquire in this experiment. Careful scrutiny of this figure raises the following issue: the Lamb shiftis defined as the energy separation between the 2S1/2 and 2P1/2 states. To measure this energydifference we need to compare two transitions where the initial states have the same energy andthe final states are the 2S1/2 and 2P1/2 states. The transitions shown in Fig. 4 do not appear tosatisfy this requirement. You will be measuring the energy difference between the 3D3/2 → 2P1/2

transition and the 3P3/2 → 2S1/2 transition, as shown in the figure. What about the energydifference between the 3D3/2 and the 3P3/2 states? This difference is 5.5 MHz—a non negligibleamount relative to the precision of Lamb shift measurements [8].

It turns out that in the hydrogen discharge tube the 3D3/2 and 3P3/2 states are broadened andmixed by the linear Stark effect. This effect is a shift of atomic energy levels in the presence ofan electric field [3, pp 219–227]. Atoms in the discharge tube experience not only the applied DCelectric field (created by the high voltage power supply), but also local time-varying fields due tothe plasma nature of the gas discharge. These strong time-varying fields smear out the Stark energyshifts thereby broadening the energy levels. When the broadened energy levels begin to overlap thetwo states can readily mix, resulting in a breakdown of the ∆l = ±1 selection rule. The broadeningand mixing result in these two states effectively sharing a common energy level, thus resolving theapparent discrepancy [9].

2 The Experiment

2.1 Saturation Spectroscopy

An inspection of Fig. 4 invites the question: how do you eliminate the Doppler broadening and getthe resolution shown at the bottom of the figure—resolution that is essential to observe the Lambshift with optical transitions? The answer: Doppler-free saturation spectroscopy. In this techniquelight with very narrow linewidth (several MHz for the diode laser used in this experiment) passesthrough a collection of hydrogen atoms undergoing Doppler-broadened transitions in a dischargetube. With its much smaller linewidth, the laser light excites (from the n = 2 state to the n = 3state) only those atoms having just the right velocity to shift them into resonance with the laserlight.

8

Figure 4: Top: Doppler broadened profile of Balmer α lines. Bottom: Balmer α lines resolved usingsaturation spectroscopy.

9

Our discharge tube is somewhat like the tube in a neon sign, only containing hydrogen instead ofneon. When it is running, the tube in the experiment contains a mixture of molecules, neutralatoms, ions (in the form of protons) and free electrons.

The frequency (or wavelength) of the laser light can be varied continuously over the entire Dopplerbroadened transition. (Indeed, it was the development around 1970 of narrow linewidth, tunablelasers that made this kind of spectroscopy possible.) If the laser frequency ν laser is set such thatνlaser < ν0, the center frequency of a particular transition, an atom will come into resonance (i.e.,will be excited from a lower energy state to a higher energy state) only if it is moving toward thelight with velocity

v =

(

ν0 − νlaser

ν0

)

c . (12)

Similarly, if νlaser > ν0 an atom will come into resonance only if it is moving away from the lightwith velocity

v =

(

νlaser − ν0

ν0

)

c . (13)

Now consider a situation where two beams of laser light are propagating in opposite directionsthrough the atoms in the discharge tube. One of the beams has considerably more power thanthe other and is called the pump beam or saturating beam. The less powerful beam is called theprobe beam. The pump beam has sufficient power to promote a significant fraction of the atomsin its path from the lower, n = 2 level to the upper n = 3 level, and if it is powerful enough, it willequalize the n = 2 and n = 3 populations, thus “saturating” the transition.

A simplified schematic of the experiment is shown in Fig. 5. Note how the two beams are shown

Figure 5: Experimental arrangement for saturation spectroscopy. The layout of optical componentsin our experiment is more complex; you should understand the function of the additional items.

to overlap (with a highly exaggerated angle of crossing) inside the hydrogen discharge tube. In ourexperiment, the two beams are more nearly parallel than the beams in Fig. 5. We shall designatethe axis along which the beams are propagating the x axis, as shown in the figure.

10

The broad bell-shaped curves in Fig. 6 depict the distribution of velocities along the x axis of thehydrogen atoms. The distribution is due to the kinetic energy of the hydrogen atoms, and is directlyrelated to the temperature of the gas. Note that the average velocity is 0—the atoms aren’t goinganywhere as a group. Those atoms with negative vx are moving toward the probe beam and awayfrom the pump beam; those with positive vx are moving toward the pump beam and away fromthe probe beam. The negative dips in the bell curves represent the atoms which absorb the laserlight at the frequency indicated; their number in turn depends on the intensity of the light and thefraction of the total which can be excited at that frequency.

The situation for νlaser < ν0 is shown in Fig. 6(a) and corresponds to the condition given in Eq. (12).The probe beam interacts with atoms moving toward it, and the pump beam interacts with atomsmoving (in the opposite direction) toward it; since the laser frequency (in the laboratory referenceframe) is less than the resonance frequency of an atom at rest, a given atom can only interactwith that laser beam which is “blue-shifted” to the resonance. Thus, each beam interacts witha different set of atoms. The opposite situation is shown in Fig. 6(c), corresponding to Eq. (13),where an atom can only interact with a beam which is “red-shifted” towards the resonance.

Consider what would happen if only the probe beam in Fig. 5 were present. The detector measuresthe light coming through the hydrogen cell. When some of the light is absorbed by the hydrogen,the detector signal will drop. When one scans the laser frequency through the resonance, theamount of absorption will mirror the bell curves shown in Fig. 6. The absorption is the convolutionof the laser intensity and the velocity distribution as a function of frequency. In other words, thesignal shows the Doppler-broadened lineshapes depicted in Fig. 4.

When the pump beam is turned on, the bell-shaped distribution of atoms which can absorb lightfrom the probe beam acquires an anomaly: the pump beam excites a portion of the atoms fromthe n = 2 to the n = 3 state, and as long as they stay in that state, they cannot absorb light fromthe probe beam. One might say that the pump beam “burns a hole” in the velocity distribution. Ifthe laser frequency is such that the hole burned by the pump beam overlaps the range of velocitiesin which atoms can absorb light from the probe beam, the result will be a slight decrease in thelight absorbed from the probe beam. One will see a slight uptick in the detector signal when thishappens. Because the two beams are directed oppositely, this situation will occur only with thoseatoms where vx = 0; this is shown in Fig. 6(b). An example of the detector output which showsthe hole-burning effect is shown in Fig. 7. In the figure the upper trace is the detector signal whichshows an overall decrease as the laser frequency is swept through the transition along with thesmall upward blips indicating hole burning.

The hole burning is seen only over the non-Doppler broadened linewidth of the transition centeredat ν0. When the laser frequency is outside this linewidth, the two beams are again interacting withdifferent sets of atoms. As the frequency of the laser is swept through the non-Doppler broadenedlinewidth centered about ν0, the probe beam intensity maps out just this linewidth—hence theterm “Doppler-free saturation spectroscopy” to describe this technique. The trace at the bottomof Fig. 4 shows a plot of the probe beam intensity as the laser frequency is swept over the entireDoppler broadened Balmer α transition.

Exercise 9 What are the relative populations of the n = 2 and n = 3 levels when they have beensaturated by the pump beam? Answer: they are equal. Why is this?

11

0

PUMP BEAMPROBE BEAM

0

(a) νlaser < ν0

(b) νlaser = ν0

(c) νlaser > ν0

0

Pump beam interacts withatoms having positive vx

Probe beam interacts withatoms having negaitive vx

vx

vx

vx

Probe and pump beamsinteract with atomshaving vx = 0

Pump beam interacts withatoms having negative vx

Probe beam interacts withatoms having positive vx

Figure 6: Absorbtion of pump and probe beams by a distribution of atoms as the laser frequencyis swept through the resonance occuring at ν0. The Bell-shaped curves show the usual Maxwellianvelocity distribution along the x axis.

12

Figure 7: Oscilloscope trace showing Doppler broadened Balmer α absorbtion line and effect ofhole burning. (Lower trace [straight line] is a control signal used to vary the laser frequency)

2.2 Lock-in detection

Figure 5 shows a fan-like device called a chopper. Where does this fit into the scheme of things? Toanswer this, first look at Fig. 7. This shows the intensity of the probe beam as the laser frequency isswept through the Doppler broadened transition. The curve is upside down relative to that shownin Fig. 4—we are looking at the remaining light after part of the beam has been absorbed by thehydrogen in the discharge tube; increasing intensity is in the upward direction.

Note the small blips in the Doppler broadened curve. These blips are due to hole burning: whenthe two beams interact with the same set of atoms, more of the probe beam intensity gets through.Note that there are 5 blips which appear to correspond to the 5 transitions shown in Fig. 4. Atthis point, you may want to claim that the measurement is over—we’ve located the various linesusing the saturation spectroscopy technique. In a sense you are right, but the signal is small, andit would be difficult to precisely locate the various lines given the small size of the blips in Fig. 7.How can we do better? Enter the lock-in amplifier. When the pump beam is turned off there is nohole burning and all we get with the probe beam is the Doppler broadened profile. By turning thepump beam on and off periodically, we can make the blips appear and disappear periodically—theperfect situation for detection with a lock-in amplifier. The pump beam is turned on and off as itis alternately passed and blocked by the chopper blades at a frequency νchop, and a signal at thisfrequency is also fed from the chopper controller (not pictured in Fig. 5) to the lock-in referenceinput. The lock-in is sensitive only to signals that appear and disappear at the reference (chopper)frequency; all it detects are the blips not the Doppler broadened curve. By plotting the lock-inoutput as the laser frequency is slowly swept through the Doppler broadened linewidth, a plotresolving the various lines of the Balmer α transition will result. This is what is actually shown at

13

the bottom of Fig. 4.

2.3 Crossover resonances

The Doppler-free saturation spectroscopy technique is clearly a very powerful tool for resolvingclosely spaced lines. It does, however, sometimes introduce extraneous features that are not partof the spectrum under investigation. Fig. 4 shows a resonance peak that looks like it might beassociated with the 3S1/2 → 2P1/2 transition, but is labeled “crossover resonance.” What is this?Note that this peak is located halfway between 3P3/2 → 2S1/2 and 3P1/2 → 2S1/2 transitions, whichshare a common lower energy level. When the laser frequency is halfway between these transitions,the pump beam saturates the higher frequency transition (3P3/2 ↔ 2S1/2) in atoms moving towardthe pump beam and saturates the lower frequency transition (3P1/2 ↔ 2S1/2) in atoms movingaway from the pump beam. The probe beam also interacts with these same sets of atoms, butbecause they are saturated by the pump beam, the atoms absorb less of the probe beam resultingin the appearance of the crossover resonance. This resonance does not affect the measurements inthis experiment, but it does have the effect of masking the weak 3S1/2 → 2P1/2 transition.

2.4 Measuring the frequency shift

The standard method of measuring the frequency shift as the laser frequency is being swept is tosend some of the laser light into a Fabry-Perot interferometer which will pass light at well definedfrequency intervals. The way this works is as follows. A Fabry-perot interferometer is made oftwo partially silvered mirrors facing each other and aligned so that their faces are exactly parallel.The mirrors reflect the light between their two inner surfaces many times. If the wavelength of thelight is such that an integer number of half-wavelengths fit in the space between the mirrors, thenthe multiple reflections allow a standing wave to be present. In other words, the mirrors form aresonant cavity for light waves, analogous to the sort of resonances one has in an organ pipe orflute for sound waves. Because the surfaces are only partially silvered, some of the light penetratesthe mirror surface and escapes out of the cavity and can be picked up by a light detector. Thedetected light intensity is largest when the resonance condition is met. If laser light sent throughthe mirrors is swept through a range of frequencies, the detector will record an intensity oscillationcorresponding to the change in frequency.

Let’s put some mathematics to this description to see how the mirrors allow us to measure thefrequency shift. Let the distance between the mirror surfaces be d, and let the laser wavelength beλ. A standing-wave resonance occurs when

2= d , (14)

where n is some integer. For light waves in a vacuum, the frequency ν and wavelength are related bythe well-known formula λν = c, where c is the speed of light. Thus the frequency of this particularstanding wave can be written

ν = nc

2d. (15)

If the frequency is increased, the wavelength will decrease, and the resonance condition will be lost,leading to a reduction in the detected light intensity. At some higher frequency, the next resonancewill come into play, as there will be n + 1 half wavelengths between the two mirrors, and the light

14

signal will peak again. If we call these two resonant frequencies ν1 and ν2, the difference betweenthem is

ν2 − ν1 = (n + 1)c

2d− n

c

2d, (16)

or∆ν =

c

2d. (17)

This expression is called the “free spectral range” of the Fabry-Perot interferometer, and it givesthe amount of frequency shift between two peaks in a detector output sweep. Note that since d isin the denominator of the right side of Eq. (17), as we increase the distance between the mirrors, weare able to resolve ever smaller frequency shifts. This means we can make spectral measurementsof very high resolution.

In our setup, the chopped pump beam is “picked off” by a mirror after it exits the dischargetube and sent through a Fabry-Perot intereferometer set at a fixed mirror spacing of about 10centimeters. A detector is placed on the other side of the interferometer to measure the lightintensity at the center of the beam. At this mirror spacing, the interferometer is highly sensitiveto vibration, especially at the resonance peaks. So we use the lock-in trick on the intereferometeras well. With a long-enough lock-in time constant, the effects of vibration can be greatly reduced,and a scan of the output signal from the lock-in amplifier shows a relatively smooth variation asthe laser frequency is swept.

You can compare the periodic variation in the Fabry-Perot signal to the peak locations in thehydrogen spectrum to calculate the frequency differences between various peaks. You will need tomake a good measurement of the mirror separation in order to get the free spectral range. Detailsof how to do this and interpret your scan are given in the data analysis section of the experimentalprocedure.

2.5 Nobel Prizes

Willis Lamb was awarded the 1955 Nobel Prize in Physics “for his discoveries concerning thestructure of the hydrogen spectrum.”

Richard Feynman, Julian Schwinger and Sin-Itiro Tomonaga were awarded the 1965 Nobel Prizein Physics “for their fundamental work in quantum electrodynamics, with deep-ploughing conse-quences for the physics of fundamental particles.”

Arthur Schawlow was awarded the 1981 Nobel Prize in Physics (along with Nicolaas Bloembergen)“for their contribution to the development of laser spectroscopy.”

References

[1] Hansch, T. W., A. L. Schawlow, and G. W. Series, “The spectrum of atomic hydrogen,”Scientific American, March 1979, pp. 94–110.

[2] R. Eisberg and R. Resnick, Quantum Physics of Atoms, Molecules, Solids, Nuclei and Parti-cles, 2e, Wiley, New York, (1985).

[3] B. Bransden and C. Joachain, Physics of Atoms and Molecules, Longman, London, (1983).

15

[4] Williams, R. C., “The fine structure of Hα and Dα under varying discharge conditions,” Phys.Rev. 54, 558 (1938).

[5] Lamb, W. E., and R. C. Retherford, “Fine structure of the hydrogen atom by a microwavemethod,” Phys. Rev. 72, 241 (1947).

[6] Welton, T. A., “Some observable effects of the quantum-mechanical fluctuations of the elec-tromagnetic field,” Phys. Rev. 74, 1157 (1948).

[7] Eides, M. I., H. Grotch and V. A. Shelyuto, “Theory of light hydrogenlike atoms,” PhysicsReports, 342, 63–261 (2001).

[8] Weber, E. W., and J. E. Goldsmith, “Double-quantum saturation spectroscopy in hydrogen:Measurement of the 3P3/2-3D3/2 Lamb shift,” Phys. Rev. Lett. 41, 940 (1978).

[9] Hansch, T. W., I. S. Shahin, A. L. Schawlow, “Optical resolution of the Lamb shift in atomichydrogen by laser saturation spectroscopy,” Nature 235, 63 (1972).

[10] Erickson, G. W., “Energy levels of one-electron atoms,” J. Phys. Chem. Ref. Data, Vol. 6, No.3, p. 831 (1977).

16

3 Experimental Procedures

In this experiment, the Lamb shift is observed in the n = 2 to n = 3 (Balmer α) transitions inatomic hydrogen. Molecular hydrogen is generated by electrolysis of water, and then dissociated intoatomic hydrogen inside a high voltage discharge tube. The discharge tube also excites the hydrogeninto the higher n states so that the Balmer α line can be explored. Tunable, monochromatic laserlight is directed through the tube from both ends, and by manipulating the counter-propagatingbeams appropriately, one can extract non-Doppler-broadened (Doppler-free) profiles of the varioustransitions, thereby resolving them and allowing for observation of the Lamb shift.

Caution: there are several hazards associated with the Lamb shiftapparatus:

1. High voltage and electrical shock.

2. High laser power (5+ milliwatts). Never stare into direct or

reflected beam!

3. Generation of hydrogen gas.

It is important to follow the written guidelines, and to handle andadjust the equipment deliberately and safely. If you have any ques-tions about how to do anything with the apparatus, consult theTA, Professor or Lab Manager.

3.1 The hydrogen generator and flow control

It is necessary for the hydrogen generator to run for a period of time to purge the tubing of anygases other than hydrogen. The TA, Professor or Lab Manager will start the hydrogen generatorand after the appropriate period of time, start the high voltage discharge. Do not attempt tostart the high voltage discharge without assistance from the TA, professor or LabManager.

Hydrogen is generated at the negative electrode in the water/KOH solution (KOH increases theconductivity of the solution, thereby reducing the amount of heat generated). This electrode isinside the plastic jar. With valve #2 closed, hydrogen has nowhere to exit the system exceptthrough the flow control valve (#3). The rate of flow through this valve is much less than therate at which hydrogen is typically generated with the result that pressure builds up and forcesthe liquid level in the jar down until it is below the electrode. At this time current through thesolution goes to zero, and hydrogen generation ceases. After some hydrogen has passed throughvalve #3, the liquid level rises to cover part of the electrode, and hydrogen generation resumes.During the course of your measurements, the current through the solution will go on and off anumber of times—this is normal operation. When hydrogen is being generated, current throughthe solution should be about 1.5–2.0 amperes.

17

During the experiment hydrogen flows through the discharge tube continuously. It is metered intoone end of the tube through valve #3, and is pumped from the other end by a mechanical pump.With the hydrogen generator running, check that the valves are as follows:

#1 open (black handle vertical)

#2 closed (black handle horizontal)

#3 Flow control valve: open to approximately 0.083” (see below)

#4 open

#5 open

#6 CLOSED! (Opening this valve would cause KOH liquid to be sucked into the discharge tubewhich could damage the apparatus.)

#7 (valve on vacuum pump) open

The hydrogen flow rate and thereby the tube pressure are very sensitive to the flow valve setting;changing the setting by one division (0.001” = one mark on the handle) may result in a pressurechange of up to 100 millitorr (pressure is sometimes stated in microns and denoted by the symbolµ; 1 µ = 1 millitorr). During operation of the discharge tube, the flow valve should be open threefull turns (0.025” per turn) plus 0.008” to 0.009” for a total opening of 0.083” to 0.084”. If thetube pressure is not in the above indicated range, change the flow valve setting by 1/2 division ata time (for example 0.083” to 0.0835”) and then wait to observe the pressure at which the tubeequilibrates.

Adjust the flow control valve (#3) so that the discharge tube pressure is in the range 200–400millitorr.

3.2 Checking the laser beam alignment

A comment about the laser: This is an expensive ($15,000), high quality instrument. Please treatit respectfully and keep the operational parameters (laser current, laser power, sweep amplitudeand frequency) within the recommended limits.

Check that the Laser Sweep switch on the switch box is in the “Off” position.

Turn on the laser controller (key switch). The controller performs a self-check, and then displaysthe value of the various parameters. After several minutes, the diode temperature should stabilizeat 20.0C, the recommended operating temperature. Check that the diode current (far right knob)is set at or near zero. Turn on the laser power (button on upper left), and when the light onthe button stops blinking, slowly turn up the diode current, noting the jump in beam intensity atabout 39 mA as the diode changes from an LED into a laser. Increase the diode current until theindicated laser power is about 4.0 mW (current should be ≈ 49 mA). Do not exceed 5.0 mWpower or 52 mA current.

With the laser on, check the alignment of the laser beams relative to each other and to the dischargetube. This is done by holding an index card up to the round holes in the wood box. First confirmthat the mechanical chopper is off, and that the chopper blade is not blocking the pump beam. Ifthe blade is blocking the pump beam, nudge it gently to a new position where it no longer blocks

18

the beam. Standing on the “laser” side of the wood box, hold an index card up to the hole on theright side of the box. Take a second index card and block the pump beam near the chopper. Nowonly the probe beam will show up on the index card at the circular hole. When the pump beamis unblocked, it should appear on the card just to the side of and at the same level as the probebeam. Now hold up the index card to the hole on the left side of the box. As before, block andunblock the pump beam. The pump beam should appear on the opposite side of the probe beamas on the right side of the box. When this geometry is obtained, the beams will cross inside thedischarge tube—a necessary condition to observe the Doppler-free line profiles.

Both the pump and probe beams should pass cleanly through the discharge tube without hittingthe walls. If the beams seem distorted or blurred out on the index cards, or if you can see thatthey are hitting the walls of the tube, consult with the TA, Professor or Lab Manager as to howto achieve the proper beam alignment.

Finally, check that the probe beam falls on the detector properly. Block the probe beam with anindex card. The detector current should decrease by approximately 10 µA. If the current does notchange by something like this amount, it may be necessary to adjust the probe beam alignmentwith mirror M5.

Exercise 10 Make a sketch of the optics and show the path of the two beams. Describe the effectof the various optical components; in particular, explain the role of the small mirror M4 and theiris. Also show the components used to measure the frequency shift.

At this time, turn on the power to the remaining electronic instruments: function generators,lock-in amplifiers (power switch on the rear of the SR830), and the scope (if they are not alreadyon).

3.3 The discharge tube

Confirm with the TA, Professor or Lab Manager that the hydrogen generator has been running forthe necessary period of time (at least an hour), and if so, arrange to have the discharge started.Do not attempt to light the discharge tube without supervision; equipment damageor personal injury may result from improper use of high-voltage equipment. When thetube is lit, confirm that the following experimental parameters are at or near their nominal values:

Discharge tube high voltage: +1900 – +2200 volts (start with +1900 V).Discharge tube pressure: 200–400 millitorr.Discharge tube current: 16–21 mA.

When all is well with the discharge, it will have a beautiful and relatively uniform magenta col-oration. Significant yellow coloration, which makes the overall discharge look whitish is a sign thatimpurities are present; such impurities will degrade the quality of the Lamb shift data. Runningthe tube for a while (≈ 1/2 hour) should flush out the impurities and result in the desired magentacoloration. If not, consult the TA, Professor or Lab Manager for advice. (Note, it is not possibleto remove all of the impurities.)

It is interesting to view the discharge tube through one of the small diffraction grating slides.If, when looking through a slide, you see three strongly colored lines—the Balmer α (red), β(turquoise) , and γ (violet) lines—with very little color between these, then the gas is fairly pure.With significant impurities, you will see other colors, mainly yellow.

19

It may happen that the light will go out. If this occurs, have the lab manager or TA restart it. Itmay be necessary to increase the high voltage or gas pressure. The lamp tends to be most unstablejust after it has been lit.

3.4 Observing Doppler-broadened lines

The two function generators put out triangular waveforms that sweep the laser wavelength throughthe transitions of interest. The Krohn-Hite puts out the fast sweep (20–30 Hz) that is used whenadjusting the system. The Stanford Research Systems DS345 puts out the slow sub-Hz sweepthat is used when acquiring data with the X-Y recorder. When the Laser Sweep is on, the centerwavelength of the sweep can be varied continuously over a small range with the Piezo Voltagecontrol.

The wavelength readout on the laser control should indicate 656.3 nm. If some other value isindicated, you probably won’t be able to see the Doppler broadened absorption lines without firstmaking a coarse adjustment of the laser wavelength, as described in the second paragraph below.If 656.3 nm is indicated, proceed as follows.

On the switchbox, set the “CH2” switch to “Sweep” and the “Sweep Speed” switch to “Fast”. Nowlook at the sweep signal on the scope. Check that the sweep frequency is approximately 40 Hz,and that the amplitude is about 3 volts peak-to-peak. (This is on the Krohn-Hite generator.)The absolute maximum sweep voltage is 5 volts peak-to-peak. Check that the “ScopeInput” switch is set to the PDA-700 position. (The PDA-700 is the little amplifier that convertsthe photodiode output current to a voltage.) Flip the “Laser Sweep” switch to the “ON” position.If you are lucky, you will see something like the lower curve in Fig. 8, where the two humps arethe Doppler-broadened absorption curves. (Beam intensity increases going upward; going through

Figure 8: Oscilloscope trace showing Doppler broadened Balmer α absorption lines. Equipmentsettings: tube pressure = 190 µ, laser power ≈ 5 mW, sweep frequency = 23 Hz, PDA-700 gain =20 µA.

20

a resonance removes energy from the beam, thus reducing the detected intensity). If you see onlya hint of these curves, vary the Piezo Voltage control to try to bring the curves onto the scopedisplay. If you are unable to get a display of the broadened absorption curves, the laser wavelengthwill need to be coarsely adjusted.

To make coarse changes in the wavelength, flip the “Laser Sweep” switch to “OFF”, set the PiezoVoltage control to 50, and push the Track button above the Wavelength Adjust knob. Changethe wavelength by turning the knob two clicks at a time. Note how the wavelength display showsresolution to 0.01 nm when the wavelength knob is turned, but only 0.1 nm otherwise. Afterchanging the coarse wavelength, turn the Track button off, the “Laser Sweep” back on and huntfor the absorption humps by varying the Piezo Voltage control. When the absorption curves aredisplayed on the scope, the wavelength readout should indicate 656.2 to 656.3 nm.

The laser sometimes exhibits an undesirable behavior known as mode-hopping, with a result shownin Fig. 9 Mode-hopping results when the cavity + diode combination jumps from one standing wavemode to the next; i.e., the number of half-wavelengths of light in the cavity changes by one. Mode

Figure 9: Oscilloscope trace showing mode hopping instability in tunable laser.

hopping usually happens when the Track button and the laser sweep signal are on simultaneously.This condition is signalled by a persistent “beeping” from the laser head, as the picomotor triesto counteract the effect of the piezo. If the mode-hopping appears and persists after the coarseadjust is turned off, varying the Piezo Voltage, reducing (or increasing) the laser power, or slightlyincreasing the laser temperature control setting will usually cause the mode hopping to go away.Please ask for help if you think you need to increase the laser power or temperature.

3.5 Locating and recording Doppler-free lines and frequency shift signal

With just one cycle of the sweep waveform and absorption curves displayed (as in Fig. 8), turn onthe AVERAGE feature of the scope (see ACQUIRE menu) and set it to 8 or so sweeps/average.

21

A close inspection of the absorption curves will reveal several small positive-going blips. At thelocation of these blips, the probe beam intensity increases (remember, increasing intensity is up onthe scope). To see the blips more clearly, increase the scope sweep speed and vertical gain as inFig. 7.

Block off the pump beam and observe what happens to the blips. Unblock the pump beam andagain observe the effect. Why does the intensity of the probe beam increase at these locations whenthe pump beam is on and decrease when the pump beam is off? It is due to the “hole burning”action of the pump beam.

When the mechanical chopper is on, it performs the same function as you did in blocking andunblocking the pump beam, i.e., it switches the Doppler-free hole burning effect on and off. Areference signal at the chopping frequency is connected to the Reference input of the lock-in, andthe output of the PDA-700 amplifier is connected to the lock-in Signal input. In this configuration,the lock-in detects the Doppler-free hole burning only, since this is the signal that is being switchedon and off at the lock-in reference (chopping) frequency.

Once you have obtained a scope display something like Fig. 7, you can proceed to record data withthe X-Y plotter as follows. This sequence of steps is designed to help you first locate the peaksclearly, using the scope, while using the lock-in amplifiers with somewhat fast time constants, andthen really getting the most out of the lock-ins by running them with very slow time constants,and concurrently, slow sweep speeds when you make the X-Y plots.

1. Check that the DS345 settings are as follows:

- Amplitude: 1.5 volts peak to peak (this is the voltage going into a 50 Ω load; the voltagewill be twice this value going into a high impedance load).

- Waveform: triangular.

- Frequency: approximately 0.05 Hz.

2. On the switch box, switch the “Scan Speed” from “Fast” to “Slow”, and turn on the “X-axis”signal to the “X-Y Plotter”. Turn on the HP X-Y plotter and set the initial sensitivities asfollows (you will probably change these):

X: 0.2 V/in, slow responseY-1: 0.1 V/in, slow responseY-2: 5 V/in, slow response

Now check that X-axis travel is reasonable—the carriage bar should cover most of the chartarea, and it should not go all the way to either limit. You may need to adjust the zero positionand the x-axis sensitivity. Since the scan voltage oscillates between a positive and negativevoltage, the X-axis zero setting should be close to the midpoint of the chart.

3. Set the oscilloscope Time/Div setting to something pretty slow, like 2s/Div. Use the horizon-tal position knob to put the display window to the beginning of the scope’s memory. (Notethe little bar indicator at the top of the scope display as you turn the position knob. Thebeginning of the memory corresponds to the right end of the bar.)

4. Turn on the mechanical chopper. The chopping/reference frequency will be displayed on theSR830 lock-in, and it should be set to around 300 Hz (the value is not critical).

5. Suggested initial settings for the SR830 lock-in are:

22

Time constant = 30 mS

12 dB/octave roll-off

Signal Input: A, DC coupled, Grounded

20 mV sensitivity

High reserve

DC coupled

Line filter ON

Channel One settings:

Display: ROutput: Display

Ratio: OffExpand: Off

Offset: Off

6. Once you have established suitable x axis travel on the X-Y recorder, reduce the sweepfrequency on the DS345 to something like 0.01 Hz. A slower frequency is necessary to allowthe lock-in to track the changes in the signal as the laser output passes through the variousresonances.

7. On the switch box, select the “Lock-in” for the “Scope Input”. Check that the “Laser Sweep”switch is on. Adjust the Volts/Div setting to comfortably show the peaks from the spectrumas the scan voltage varies. Have patience. You should be able to see a big peak followedby a space and then four smaller peaks (or the other way around, depending on the scandirection).

8. On the switch box, select the “F-P Signal” for the “CH2” input. The initial settings on theEG&G 5210 lock-in (the one with the black face panel) should be as follows:

Sensitivity

Input: V, AFloat/Ground switch: Out (grounded)

Sens. setting: 100 mV

FiltersMode: BP

Set Freq: TrackLine reject: “F” and “2F” both on

Display 1

Ref F: selectedReference: “Int” and “2F” both off

Display 2

Time constant 30 mSSlope: 12 dB

Dyn Res: NormDisplay: RΘ

23

9. Turn on the PDA-700 amplifier that is connected to the Fabry-Perot interferometer detector(if not already on). Set the “Range” switch to .xxxx µA. Cover the entrance of the detectorpinhole with your finger and adjust the “Offset” knob so that the display reads zero (or asclose as you can make it). Release your finger, and note the change in current when the laserlight hits the detector.

10. Now you should be seeing two kinds of traces on the oscilloscope: A CH1 trace that shows theoutput of the probe-beam signal as processed by the SR830 lock-in amp, with the Doppler-free absorption peaks, and a CH2 trace that shows the output of the interferometer detector,with the periodic peaks showing the frequency shift of the laser.

If you don’t see these traces clearly, try adjusting the scope controls and/or lock-in sensitivitylevels. (At the time constant settings chosen, there will be a fair amount of noise.) If you stillcan’t get the traces to appear properly, turn off the chopper power (make sure that after theblade stops it is not blocking the beam), turn off the X-Y plotter switches, and switchthe “Sweep speed” back to the fast sweep and verify that the blips due to hole burning arepresent on the Doppler-broadened lines displayed on the scope.

11. On the Switch Box, select the “Lock-in” for the “Plotter Y-axis” input, check that the “LaserSweep” is on, and turn the “Y-axis” switch for the “X-Y plotter” on. Do not put any pensin the plotter yet. Watch the pen carriages move as the carriage bar moves back and forth.Note the extent of the carriage travel. Reduce or increase the sensitivity on each channel, ifnecessary, to insure that a good plot will result.

12. Turn the “Servo” switch on the recorder to “Standby”, the “Pen” switch to “Lift” and the“Chart” switch to “Release”. Carefully slide a piece of graph paper onto the surface of therecorder and insure that the bottom edge is flush against the metal rail. Flip the “Chart”switch to “Hold” and smooth the paper with your hand; it should stay snug on the surface.

Now (and not before) remove the pens from their plastic box. Uncap and put onein each carriage. Note that the pens have an angled tip: the pen for the outer carriage shouldbe placed in the holder so that the tip points in towards the carriage bar, and the pen forthe inner carriage should be placed so that the tip points away from the carriage bar. Thisinsures that the horizontal displacement between the two traces is minimized. Make sure thatthe plastic extension on each pen is securely held by the carriage clip.

13. After the pens have been secured, turn the “Servo” switch “On”.

14. On the DS345 function generator, set the frequency to 0.003 Hz, and then set the timeconstants on both lock-in amps to 300 ms.

15. Wait until the carriage bar reaches the end of its travel, and then flip the “Pen” switch on theX-Y recorder to “Record”. Hint: you can reduce your waiting time by setting the functiongenerator to, say 0.1 Hz to make it reach the beginning point of your scan more quickly, andthen load in “.003” (or slower) to the generator’s buffer, but wait to hit the “Hz” button untilright before your scan starts.

16. Once you have recorded a set of the Balmer-α lines, you can fine-tune the various settings tooptimize the data. In particular you should try to maximize the use of the horizontal chartarea, since that is the dimension you will use for the data analysis. If the peaks are closetogether on the chart, you will lose precision. To record the lines as accurately as possible, itmay be necessary to decrease the sweep speed to something like 0.002–0.001 Hz.

24

17. When you are happy with the settings for your scans, record two charts: one with the scanmoving from left to right, and another with the scan moving from right to left. You willnotice that these two directions do not line up! Notice the frequency shift trace. You mayalso see that the peak spacing is not uniform. This effect and how to deal with it is discussedin the analysis section. Do not touch anything on the optics table while your scans

are running! If you do, you will disturb the Fabry-Perot mirrors and cause aglitch in your frequency shift scan.

18. After taking your scans, immediately set the “Servo” switch to “Standby” and re-

move and recap the pens! The pens for this recorder are expensive ($20 each) and hardto obtain. They also dry out quickly since they have small reservoirs.

3.6 Shut Down

1. Remove the pens from the chart recorder, recap them, and return them to the plastic box.(Yes, I did just say this!)

2. Turn down the high voltage setting to 0 volts. Flip the High Voltage switch to the Stand-by/Reset position. Turn off the high voltage power supply.

3. Close valve #4 and open valve #2 so the hydrogen generator can vent to the room.

4. Turn off the hydrogen generator power supply.

5. Turn the laser diode current down to 0, turn off the laser power (push button), and turn offthe laser controller (key).

6. Turn off the remaining electronic instruments.

7. After the discharge tube has pumped down to the 20–30 millitorr range, close valves #5 and#7, and unplug/switch off the vacuum pump (depending on your model).

8. Cover up the optics.

9. Make sure that the pens are capped and put away.

4 Measure the mirror spacing

In order to relate the peaks in the interferometer scan to the frequency shift, you must know the freespectral range of the interferometer, and for this, you need to know the spacing d of the Fabry-Perotmirrors.

Use a caliper to measure the distance between the inside surfaces of the aluminum mirror mountsby placing the caliper between the mounting flanges at the edge. Be careful to make the caliperperpendicular to the flange surfaces. You should measure the distance between three differentpoints and then average the results. If you don’t understand these instructions, ask for help! Notethe variation in the measurements and calculate an uncertainty.

Also, write down the amount of recess that each mirror surface has relative to the surface of themount. These numbers are written on the sides of each aluminum mounting flange.

25

5 Data Analysis

To complete the data analysis, you should have two X-Y plots showing the peaks of the Balmer-αline—one for each scan direction, and a good measurement of the Fabry-Perot mirror spacing.

First, use a ruler and a sharp pencil to mark the centers of all of the peaks in your plots. Thepeaks from the interferometer are relatively broad, so you will need to use your best judgment toestimate their centers. Discuss any choices that might seem questionable in your report.

Next, use a caliper to measure the distance between successive peaks to within a tenth of a mil-limeter or a hundredth of an inch (depending on your choice of paper units).

Before proceeding further, compare the peak spacing between similar peaks for the two separateplots. In particular, note any regular trends in the peak-spacing of the interferometer scans. Howmuch change is there from the beginning of the scan to the end? How much change is there inthe region of the hydrogen peaks? You should think about, and discuss with your partners, whatthe best way would be to get the highest accuracy in your measurement of the spacing of the two“Lamb shift peaks”: the 2S1/2 → 3P3/2 and 2P1/2 → 3D3/2 transitions. For example, should youuse the average interferometer peak spacing over the whole scan, the average over just the rangethat includes the hydrogen peaks, or just the one or two interferometer peaks near the Lamb shift?Whatever method you decide to use, justify it in your report. Calculate the Lamb shift for bothof your scans and compare them to each other. Do they agree within the uncertainty you wouldcalculate? Average your two results and compare the average to the accepted value given in thetheoretical discussion of the effect. Discuss.

The systematic variation in interferometer peak spacing is due to hysteresis in the piezo-electricresponse of the laser scanning mechanism. The piezo crystal flexes under the influence of an electricfield. This flexing is what moves one of the mirrors in the laser cavity, which then changes theresonant frequency and hence the color of the laser. The hysteresis is a “memory” of the crystal’sprevious state, e.g., if the crystal was in a compressed state, it tends to stay that way even if theapplied voltage is changed so that it should expand. As a consequence, the crystal flexes less atthe beginning of the scan than at the end, and the interferometer peaks are more widely spacedat the beginning than at the end. In high-precision work, the interferometer is output is coupledto the laser and to an independent frequency standard in order to counteract this hysteresis. Oursetup does not have this added control feature, but you should be able to obtain the Lamb shift towithin a few percent.

You should also calculate the separation between the peaks at the extreme ends of the spectrum,and compare your result to the accepted value:

∆ν = 9890 MHz between the 2P3/2 → 3D5/2 transition and the 2P1/2 → 3D3/2 transi-tion.

This is derived by comparing published values for frequency intervals between the 3D states andbetween the 2P states [10].

Prepared by J. Stoltenberg, D. Pengra, and R. Van Dycklamb_shift.tex -- Updated 14 February 2008

26